Hi
How come i overlooked this post for so long?
The steps are easy enough to outline:
1 - first layer intuitively, it's not really necessary to solve all centers
2 - middle layers edges, for 5x5x5 make sure u align the very middle correctly (never a quarter turn away from solved) Make use of the freedom u have to rotate the middle layers freely to solve edges with *no algs* before using algs to solve remaining ones. (I hold the solved first layer on the left.)
3 - last layer corners, permute then orient. Or learn CLL or even COLL if u can see how to use that to make edges easier. I don't use CLL/COLL.
4 - last layer edges. I solve them by *freecycles* for those who know them from blindsolving. I make them up to solve 2 edges at a time whenever possible. I do NOT orient then permute. That's a waste. Direct solving is what it's about!
(5 parity) - i only use one alg for that. Conceptually i put the 2 edges to be *swapped* into the same layer, the turn that layer a quarter turn and fix the layer with an edge 3-cycle. Putting 2 edges into same layer (i use b-layer) never takes more than 2 turns plus a cube orientation. Undoing the setup at the end is trivial.
6 - the rest of the centers. More commutator cycling orgy
Make use of block cycles as much as u can! Also use 3x3x3-style cycles when u see that they are productive. Cage method takes more turns than pairing-up. But personally i find that recognition is much better and i have very few delays in my solves when im fully focused.
I don't really want to give out any algorithms yet. They are almost trivially easy and very fast. For last layer edge 3-cycle think about how u can make setup moves to bring one edge to the D-layer while leaving 2 in the same *orbital* in the U-layer. Once that is done, always 1 turn only u can make easy 3-cycle commutator. Recognition becomes easy with practice <_<
-Per
PS! Moderator can u disable that capital b (B)) turns into an emoticon??